Our objective is to determine the extent to which different aspects of defense mechanisms and conditioned fear are relevant to features of particular forms of normal and abnormal anxiety in humans. We make a distinction between two forms of aversive states, fear and anxiety, with a particular focus on anxiety. Operationally, fear is defined as a phasic aversive state induced by a proximal and identifiable threat, whereas anxiety is a more sustained form of response focused on potential distal or future threats that are not clearly identifiable. Research in animals suggests that these two forms of aversive states can be modeled by cue and context conditioning, respectively. Experimental models: We focus on developing models of anxiety based on their presumed clinical relevance. In the last several years, we used virtual reality to model cue and context conditioning. In this experiment, fear and anxiety are assessed based on physiological responses. We recently introduced a new experiment, a virtual reality analogue of the Morris water maze. In this task, anxiety is measured based on behavior. The Morris water task also presents important procedural differences compared to context conditioning. The shock is uncontrollable in context conditioning, whereas stressor duration is contingent on subjects behavior (i.e., reaching the hidden platform) in the Morris water maze. Psychopharmacology: We have demonstrated that contextual anxiety is increased by acute treatment with the serotoninergic reuptake inhibitor (SSRI), citalopram, and reduced by 2-week treatment with the same drug. These results are consistent with the effects of SSRI treatment in patients. The anxiolytic effects of SSRI are dependent on a chronic administration of the drug. Initially, SSRI can be anxiogenic in some patients. We recently examined the role of cortisol in fear and anxiety. Acute hydrocortisone treatment was found to be anxiolytic. These results are significant because cortisol's effect on anxiety is believed to be mediated via its action on the consolidation and retrieval of emotional memories. Our results suggest a direct effect on the expression of anxiety. One possibility is that cortisol potentiates the effect of the stress hormone corticotrophin releasing factor in limbic system structures such as the bed nucleus of the stria terminalis. We also found that 4-day treatment is anxiolytic. This effect was unexpected, but is not inconsistent with observation that short-term treatments with cortisol can improve mood. Behavioral studies in patients: We showed that individuals with panic disorders have elevated contextual anxiety, but normal cued fear. We recently reported a similar pattern of responses in Posttraumatic stress disorder but not in social phobia. These results suggest that patients with panic disorder and PTSD may share a common vulnerability to diffuse contextual threats. They also suggest that anxious patients are overly sensitive to unpredictable stressors. While patients with social anxiety disorder are not overly sensitive to non-specific stressors (e.g. mild shocks), they tend to form aversive Pavlovian associations with social stressors, suggesting that conditioning may be an etiological factor in this disorder. Neuroimaging: Using functional Magnetic Resonance Imaging, we demonstrated that predictable and unpredictable aversive events activate common and distinct brain areas. Notably, the amygdala and anterior insula is activated by both types of threats, whereas the bed nucleus of the stria terminalis is activated uniquely during unpredictable shocks, probably because unpredictable shocks evoke a sustained state of anxiety. In contrast, the subgenual anterior cingulate is activated during the predictable condition. This structure has been shown to mediate inhibition of fear in a number of procedures in humans and animals. Because predictable shocks are less stressful than unpredictable shocks, the subgenual anterior cingulate may be involved in reducing anxiety in the predictable condition compared to the unpredictable condition. We are also studying how anxiety affects cognitive-attentional processes. Defining cognitive-attentional changes that are central to anxiety may provide important links between brain dysfunction and clinical phenomenology. For example, we examined the effect of unsignaled shocks on reflexive vs. controlled oculo-motor responses (i.e., pro- vs. anti-saccades). Our findings revealed that anxiety promoted reflexive responding at the expense of more controlled responding.